Studies of an Enneanuclear Manganese Single

Published on Web 03/25/2005
Studies of an Enneanuclear Manganese Single-Molecule
Magnet
Stergios Piligkos,† Gopalan Rajaraman,† Monica Soler,‡ Nadeschda Kirchner,§
Joris van Slageren,§ Roland Bircher,¶ Simon Parsons,# Hans-Ulrich Güdel,¶
Jens Kortus,| Wolfgang Wernsdorfer,⊥ George Christou,‡ and Euan K. Brechin*,†,#
Contribution from the School of Chemistry, The UniVersity of Edinburgh, West Mains Road,
Edinburgh, EH9 3JJ, U.K., Department of Chemistry, The UniVersity of Manchester,
Oxford Road, Manchester, M13 9PL, U.K., Department of Chemistry, UniVersity of Florida,
GainesVille, Florida 32611-7200, Laboratoire Louis Néel-CNRS, 38042 Grenoble,
Cedex 9, France, Department of Chemistry and Biochemistry, The UniVersity of Bern,
Freiestrasse 3, CH-3000, Berne 9, Switzerland, 1 Physikalisches Institut, UniVersitat Stuttgart,
Pfaffenwaldring 57, 70550 Stuttgart, Germany, MPI für Festkörperforschung, Heisenbergstr. 1,
D-70569 Stuttgart, Germany, and IPCMS - GMI, 23, rue du Loess, F-67034 Strasbourg, France
Received December 22, 2004; E-mail: [email protected]
Abstract: The reaction of [Mn3O(O2CMe)6(py)3] with the tripodal ligand H3thme (1,1,1-tris(hydroxymethyl)ethane) affords the enneanuclear complex [Mn9O7(O2CCH3)11(thme)(py)3(H2O)2] 1‚1MeCN‚1Et2O. The
metallic skeleton of complex 1 comprises a series of 10 edge-sharing triangles that describes part of an
idealized icosahedron. Variable temperature direct current (dc) magnetic susceptibility data collected in
the 1.8-300 K temperature range and in fields up to 5.5 T were fitted to give a spin ground state of S )
17/ with an axial zero-field splitting parameter D ) -0.29 cm-1. Ac susceptibility studies indicate frequency2
dependent out-of-phase signals below 4 K and an effective barrier for the relaxation of the magnetization
of Ueff ) 27 K. Magnetic measurements of single crystals of 1 at low temperature show time- and
temperature-dependent hysteresis loops which contain steps at regular intervals of field. Inelastic neutron
scattering (INS) studies on complex 1 confirm the S ) 17/2 ground state and analysis of the INS transitions
within the zero-field split ground state leads to determination of the axial anisotropy, D ) -0.249 cm-1,
and the crystal field parameter, B40 ) 7(4) × 10-6 cm-1. Frequency domain magnetic resonance
spectroscopy (FDMRS) determined the same parameters as D ) -0.247 cm-1 and B40 ) 4.6 × 10-6
cm-1. DFT calculations are fully consistent with the experimental findings of two Mn(II) and four Mn(III)
ions “spin up” and three Mn(IV) ions “spin down” resulting in the S ) 17/2 spin ground state of the molecule,
with D ) -0.23 cm-1 and U ) 26.2 K.
Introduction
Polymetallic clusters of paramagnetic transition metal ions
can display the phenomenon of single-molecule magnetism
(SMM) if they combine a relatively large spin ground state (S)
with a large and negative (easy-axis-type) magnetoanisotropy
(D).1 This is a rare combination which results in a significant
barrier to thermally activated magnetization relaxation, and thus
far, is most often found in complexes containing manganese.
Compared to other transition metals clusters of manganese are
#
The University of Edinburgh.
The University of Manchester.
‡ The University of Florida.
⊥ Laboratoire Louis Néel.
¶ University of Bern.
§ University of Stuttgart.
| MPI für Festkörperforschung.
†
(1) (a) Sessoli, R.; Gatteschi, D.; Hendrickson, D. N.; Christou, G. MRS Bull.
2000, 25, 66-71. (b) Sessoli, R.; Tsai, H.-L.; Schake, A. R.; Wang, S.;
Vincent, J. B.; Folting, K.; Gatteschi, D.; Christou, G.; Hendrickson, D.
N. J. Am. Chem. Soc. 1993, 115, 1804-1816. (c) Sessoli, R.; Gatteschi,
D.; Caneschi, A.; Novak, M. A. Nature 1993, 365, 141-143. (d) Caneschi,
A.; Gatteschi, D.; Sessoli, R.; Barra, A. L.; Brunel, L. C.; Guillot, M. J.
Am. Chem. Soc. 1991, 113, 5873-5874.
5572
9
J. AM. CHEM. SOC. 2005, 127, 5572-5580
often characterized by large spin ground states, and this in
conjunction with the presence of Jahn-Teller distorted Mn3+
ions makes manganese clusters ideal candidates for SMMs.2
The most studied family of SMMs are the dodecanuclear
complexes of formula [Mn12O12(O2CR)16(H2O)x]n- (n ) 0, 1,
2; x ) 3, 4), but manganese clusters ranging in nuclearity from
2-84 have displayed SMM behavior.3 Because these molecules
not only display magnetization hysteresis but also quantum
tunneling of magnetization (QTM),4 they are promising new
(2) Hendrickson, D. N.; Christou, G.; Ishimoto, H.; Yoo, J.; Brechin, E. K.;
Yamaguchi, A.; Rumberger, E. M.; Aubin, S. M. J.; Sun, Z.; Aromi, G.
Mol. Cryst. Liq. Cryst. 2002, 376, 301-313.
(3) See for example: (a) Tasiopoulos, A. J.; Vinslava, A.; Wernsdorfer, W.;
Abboud, K. A.; Christou, G. Angew. Chem., Int. Ed. 2004, 43, 21172121. (b) Miyasaka, H.; Clérac, R.; Wernsdorfer, W.; Lecren, L.; Bonhomme, C.; Sugiura, K.; Yamashita, M. Angew. Chem., Int. Ed. 2004, 43,
2801-2805. (c) Murugesu, M.; Wernsdorfer, W.; Raftery, J.; Christou, G.;
Brechin, E. K. Inorg. Chem. 2004, 43, 4203-4209. (d) Milios, C. J.;
Raptopoulou, C. P.; Terzis, A.; Lloret, F.; Vicente, R.; Perlepes, S. P.;
Escuer, A. Angew. Chem., Int. Ed. 2003, 43, 210-212. (e) Wittick. L. M.;
Murray, K. S.; Moubaraki, B.; Batten, S. R.; Spiccia, L.; Berry, K. J. Dalton
Trans. 2004, 1003-1011.
(4) Friedman, J. R.; Sarachik, M. P.; Tejada, J.; Ziolo, R. Phys. ReV. Lett.
1996, 76, 3830-3833.
10.1021/ja042302x CCC: $30.25 © 2005 American Chemical Society
Enneanuclear Manganese Single-Molecule Magnet
Scheme 1. Tripodal Ligands H3thme (left), H3tmp, and H4peol
(right)
materials for data storage and quantum computing and, as such,
are the focus of much attention.
To make clusters containing Mn3+ and/or Mn4+ ions, it is
necessary to either oxidize simple Mn2+ salts in the presence
of flexible organic bridging ligands, such as carboxylates,
β-diketonates, and alkoxides, or to use the preformed metal
triangles of formula [Mn3O(O2CR)L3]0/+. This last strategy has
proved to be an extremely successful method.5 The use of these
triangles overcomes the lack of a suitable and simple Mn3+
precursor, although recently an SMM derived directly MnF3
has been reported.6 In an effort to produce large clusters with
high spin ground states and possible SMM behavior, we have
been exploring the reactivity of the tripodal ligand 1,1,1-tris(hydroxymethyl)ethane and its analogues (Scheme 1) with oxocentered manganese triangles. Previously these types of ligands
have been used to great extent in the synthesis of oxomolybdenum and oxovanadium clusters, but it is only recently that
their use has been extended to the synthesis of paramagnetic
clusters of first-row transition metals.7,8 We recently reported
the syntheses, structures, and magnetic properties of a series of
manganese rodlike complexes whose structures are based on a
series of edge-sharing triangles and whose topology could be
controlled by variation of carboxylate, tripod, and base.9 We
now report an extension to this methodology in the preparation
and characterization of an enneanuclear manganese complex.
Experimental Section
Syntheses. All reactions were carried out in aerobic conditions using
materials as received. The triangular species [Mn3O(O2CCH3)6(py)3]
was made as previously described.
[Mn9O7(O2CCH3)11(thme)(py)3(H2O)2] (1). [Mn3O(O2CCH3)6(py)3]
(0.50 g, 1 equiv) and H3thme (1 equiv) were stirred in MeCN (30 mL)
for 12 h. The solution was then filtered and layered with Et2O. Black
crystals suitable for crystallography formed in 1 week. Anal. Calcd
(found) for Mn9C42H61N3O34: C, 30.64 (30.69); H, 3.73 (3.59); N, 2.55
(2.65); Mn, 30.03 (30.22).
X-ray Crystallography. Diffraction data were collected with Mo
KR X-radiation (λ ) 0.71073 Å) on a Bruker Smart APEX diffractometer equipped with an Oxford Cryosystems low-temperature device
operating at 150 K. An absorption correction was applied using the
SADABS program;10 the structure was solved using Patterson methods
(5) Sañudo, E. C. et al. Polyhedron 2003, 22, 2267-2271.
(6) Jones, L. F.; Rajaraman, G.; Brockman, J.; Murugesu, M.; Raftery, J.; Teat,
S. J.; Wernsdorfer, W.; Christou, G.; Brechin, E. K.; Collison, D. Chem.s
Eur. J. 2004, 10, 5180-5194.
(7) (a) Khan, M. I.; Zubieta, J. Prog. Inorg. Chem. 1995, 43, 1-149. (b)
Cavaluzzo, M.; Chen, Q.; Zubieta, J. J. Chem. Soc., Chem. Commun. 1993,
131-132. (c) Finn, R. C.; Zubieta, J. J. Cluster Sci. 2000, 11, 461-482.
(8) (a) Jones, L. F.; Batsanov, A.; Brechin, E. K.; Collison, D.; Helliwell, M.;
Mallah, T.; McInnes, E. J. L.; Piligkos, S. Angew. Chem., Int. Ed. 2002,
41, 4318-4321. (b) Brechin, E. K.; Soler, M.; Christou, G.; Helliwell, M.;
Teat, S. J.; Wernsdorfer, W. Chem. Commun. 2003, 1276-1277. (c)
Moragues-Cánovas, M.; Rivière, E.; Ricard, L.; Paulsen, C.; Wernsdorfer,
W.; Brechin, E. K.; Mallah, T. AdV. Mater. 2004, 13, 1101-1105. (d) Shaw,
R. et al. Chem. Commun. 2004, 1418-1419.
(9) Rajaraman, G.; Murugesu, M.; Soler, M.; Wernsdorfer, W.; Helliwell, M.;
Teat, S. J.; Christou, G.; Brechin, E. K. J. Am. Chem. Soc. 2004, 126,
15445-15457.
ARTICLES
(DIRDIF)11 and refined by full-matrix least squares against F2 (Shelxtl)10
using all unique data. H-atoms were treated either as part of a rotating
rigid group or with a riding model. A cluster of peaks in the Fourier
difference map in the region of the hexagon of oxygen atoms attached
to Mn4-9 were interpreted as a 3-fold disordered set of two acetates
and two coordinated water molecules. The acetates were restrained to
be geometrically similar; the major component C-atoms were refined
with anisotropic thermal parameters, the minor component only
isotropically. Disordered solvent regions were treated in the manner
described by van der Sluis and Spek.11 This treated 276 e per cell or
69 e per formula unit. This fits quite well with 1MeCN and 1Et2O (66
e) for formula unit. The values of F(000), the density, µ, etc. reflect
this assumption.
Other Measurements. Elemental analyses were performed by The
University of Manchester Microanalytical Service.
Variable temperature magnetic susceptibility (dc and ac) measurements were made on a Quantum Design MPMS-XL SQUID magnetometer equipped with a 7 T magnet. Data were collected on powdered
samples restrained in eicosane to prevent torquing. Diamagnetic
corrections were applied using Pascal’s constants. Magnetization versus
field hysteresis and direct current (dc) decay measurements at temperatures below 1.8 K were performed on single crystals using an array
of home-built micro-SQUIDs equipped with three orthogonal field coils,
which allows the applied magnetic field to be turned in all directions.12
The INS experiments were performed on the inverted geometry timeof-flight spectrometer IRIS at the pulsed neutron spallation source ISIS
at the Rutherford Appleton Laboratory, U.K., using the PG002 graphite
analyzer with an analyzing energy of 1.84 meV. Data were collected
at three temperatures (1.6, 12, and 20 K) and corrected for the background and detector efficiency by means of empty cell measurements
and a vanadium reference. The time-of-flight to energy conversion and
the data reduction were done using the GUIDE program. The instrumental resolution derived from a vanadium metal reference at the elastic
position was 18 µeV. On IRIS, the ZnS detector banks cover the angular
range 2θ ) 20-160°, giving access to a momentum transfer range Q
) 0.3-1.8 Å-1 at the elastic position. A 2.5 g fresh polycrystalline
sample of undeuterated [Mn9O7(OAc)11(thme)(py)3(H2O)2] was used.
The sample was placed under helium in an aluminum hollow cylinder
can with an outer diameter of 23 mm and a sample thickness of 2 mm.
The container was inserted in a standard ILL orange cryostat.
FDMRS spectra were recorded on a polycrystalline powder pellet
(mass ) 179.16 mg, thickness ) 0.1456 cm; diameter ) 1 cm) using
linearly polarized radiation in the frequency range of 2.8-5.0 cm-1 at
temperatures of 1.8-30 K in the absence of an external magnetic field.
The spectrometer has been described in the literature.13,14
The Naval Research Laboratory Molecular Orbital Library (NRLMOL) program package is an all-electron implementation of densityfunctional theory (DFT). NRLMOL combines large Gaussian orbital
basis sets, numerically precise variational integration, and an analytical
solution of Poisson’s equation in order to accurately determine the selfconsistent potentials, secular matrix, total energies, and HellmannFeynman-Pulay forces.15 Starting from the X-ray structure we
generated an isolated single molecule for the calculation. To reduce
the computational cost we replaced the outer methyl groups with
(10) Sheldrick, G. M. SHELXTL, version 5; Bruker AXS: Madison, Wisconsin,
1995.
(11) (a) van der Sluis, P.; Spek, A. L. Acta Crystallogr., Sect. A 1990, 46, 194201. (b) Spek, A. L. PLATON, A Multipurpose Crystallographic Tool;
Utrecht University: Utrecht, The Netherlands, 1998. (c) Incorporated into
the WinGX suite: Farrugia, L. J. J. Appl. Crystallogr. 1999, 32, 837-838.
(12) Wernsdorfer, W. AdV. Chem. Phys. 2001, 118, 99-190.
(13) van Slageren, J. et al. Phys. Chem. Chem. Phys. 2003, 5, 3837-3843.
(14) Kozlov, G. V.; Volkov, A. A. In Millimeter and Submillimeter WaVe
Spectroscopy of Solids; Grüner, G., Ed.; Springer: Berlin, 1998.
(15) (a) Pederson, M. R.; Jackson, K. A. Phys. ReV. B 1990, 41, 7453-7461.
(b) Jackson, K. A.; Pederson, M. R. Phys. ReV. B 1990, 42, 3276-3281.
(c) Porezag, D. V.; Pederson, M. R. Phys. ReV. A 1999, 60, 2840-2847.
(d) Pederson, M. R.; Porezag, D. V.; Kortus, J.; Patton, D. C. Phys. Status
Solidi B 2000, 217, 197-218.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005 5573
Piligkos et al.
ARTICLES
hydrogen leaving the magnetic core unchanged. This resulted in a cluster
of 113 atoms used for the DFT calculation. The basis set for each Mn
contained 11 s-like, 5 p-like, and 4 d-like contracted Gaussian orbitals
with 20 exponents ranging from 0.04-3.6 × 106. The oxygen atoms
were described by 8 s-, 4 p-, and 3 d-contracted orbitals using 13 bare
Gaussians between 0.105 and 6.12 × 104. Nitrogen and carbon had
the same number of contracted orbitals, with 13 bare Gaussians between
0.094 and 5.17 × 104 for N and 12 bare Gaussians between 0.077 and
2.22 × 104 for C. Finally, hydrogen had 5 s-like, 3 p-like, and 1 d-like
orbitals with 6 exponents between 0.074 and 77.84. For all orbitals
not corresponding to an atomic orbital we used the longest ranged single
Gaussians. This basis set resulted in a total number of 3494 orbitals
for the complete cluster. The calculations have been carried out using
the generalized gradient approximation of Perdew, Burke, and Ernzerhof16 to the exchange and correlation functional.
The magnetic anisotropy energy, which is the key property of any
single-molecule magnet, arises mainly because of spin-orbit coupling
and other relativistic terms in the Hamiltonian.17 Pederson and Khanna
have recently developed a method accounting for second-order
contributions to the magnetic anisotropy energies.18 This method relies
on a simple albeit exact method for spin-orbit coupling and a secondorder perturbative treatment of the spin Hamiltonian to determine the
dependence of the total energy on spin projection and is implemented
in the NRLMOL package. This method has been generalized to arbitrary
symmetry and applied successfully to predict the magnetic anisotropy
energies of several single-molecule magnets.19
Results and Discussion
Syntheses. Reaction of 1 equiv of H3thme with the neutral
triangle [Mn3O(O2CMe)6(py)3] in MeCN leads to the formation
of a dark brown/black solution and a light brown precipitate.
Removal of the precipitate and subsequent layering of the
brown/black solutions with Et2O leads to the formation of
crystalline [Mn9O7(O2CR)11(thme)(py)3(H2O)2] (1) after approximately 1 week. The yield is variable, but easily reproducible, varying from a minimum of approximately 15% to a
maximum of 25%. As with many reactions involving manganese, there is likely to be many species in equilibrium in
solution, with 1 being the least soluble. Indeed, the solution
remains quite darkly colored after the crystallization of the
products is complete. The fact that complex 1 is made from a
mixed-valent triangle containing one Mn2+ ion and two Mn3+
ions and fully protonated tripodal ligands and is itself mixedvalent but containing two Mn2+, four Mn3+, and three Mn4+
ions means that its formation is a complicated mechanism
involving deprotonation/protonation, structural rearrangements,
and redox chemistry of many species present in solution.
It is interesting to note that the equivalent reactions to that
which produces 1, but using the triangles [Mn3O(O2CR)6(py)3]
where R ) Ph or C(CH3)3 with the tripodal ligand H3thme,
afford the dodecanuclear and octanuclear manganese rodlike
complexes [Mn12O4(OH)2(O2CPh)12(thme)4(py)2] 2 and [Mn8O4(O2C(CH3)3)10(thme)2(py)2] 3, respectively.9 These compounds
have metal topologies that are dominated by the presence of
the tripodal ligands, not the carboxylates. The disposition of
the three alkoxide arms of the trianion directs the formation of
triangular M3 units, where each arm of the ligand bridges one
(16) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 38653868.
(17) van Vleck, J. Phys. ReV. 1937, 52, 1178-1198.
(18) (a) Pederson, M. R.; Khanna, S. N. Phys. ReV. B 1999, 59, 693-696R. (b)
Pederson, M. R.; Khanna, S. N. Phys. ReV. B 1999, 60, 9566-9572.
(19) Kortus, J.; Pederson, M. R.; Baruah, T.; Bernstein, N.; Hellberg, C. S.
Polyhedron 2003, 22, 1871-1876.
5574 J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005
Figure 1. Structure of complex 1 in the crystal viewed from (a) above the
central [MnIII4MnII2O6]4+ ring; (b) perpendicular to the central [MnIII4MnII2O6]4+ ring; (c) the metal-oxygen core viewed from above the central
[MnIII4MnII2O6]4+ ring; (d) an idealized icosahedron showing the positions
of the Mn ions in 1 (pink) and the missing vertexes (blue); (e) binding
mode of thme3-.
edge of the triangle. These triangular units then combine in an
edge-sharing fashion to form complexes whose structures are
based on rods. The length of the rod formed is dependent upon
the identity of both the tripod and the carboxylates that occupy
the vacant coordination sites. None of these rodlike complexes
are formed from acetate containing manganese triangles, unless
a separate capping ligand, such as triethanolamine, is added.
This is most likely due to the solubility of the products
containing acetate. Complexes 2 and 3 form quickly in reasonable yields directly from slowly evaporating solutions of MeCN.
Complex 1 is not isolated as easily and does not appear from
the same slow evaporation. It instead requires the diffusion of
diethyl ether into the MeCN solutions and crystallization periods
of up to 1 week.
Description of Structure. Labeled plots of 1 are given in
Figure 1, and selected bond lengths and angles are given in
Table 2. The metallic skeleton comprises a series of 10 edgesharing triangles that describe part of an idealized icosahedron,
in which 3 of the 12 vertexes are missing. The metal ions are
centrally connected via a combination of seven oxides and three
alkoxides. The [MnIV3MnIII4MnII2O7]14+ central core can be
described as either a series of vertex- and edge-sharing [Mn3O]
units or perhaps most easily as a [MnIII4MnII2O6]4+ ring on
which is sitting a smaller [MnIV3O]10+ ring. This [MnIII4MnII2O]4+ ring is trapped-valence with the Mn2+ ions being Mn4
and Mn8. The upper [Mn3O]10+ ring consists of only Mn4+ ions
(Mn1-Mn3) and is capped on one side by the sole tripodal
ligand, each arm of which (O1T-O3T) acts as a µ2-bridge and
on the other side by a µ3-oxide (O1). The [MnIV3O]10+ unit is
connected to the [MnIII4MnII2O]4+ ring via the six µ3-oxides
within the six-membered wheel and three carboxylates which
bridge in their familiar syn, syn, µ-mode. The remaining
carboxylates in the structure bridge between the Mn ions in the
Enneanuclear Manganese Single-Molecule Magnet
ARTICLES
In the crystal, the molecules of complex 1 pack in two
directions, which are approximately perpendicular to each other,
in a head-to-head and tail-to-tail fashion, meaning that there
are two, approximately transverse, directions for the easy-axis
of magnetization (vide infra).
Magnetochemistry. DC Magnetic Susceptibility. Variable
temperature magnetic susceptibility data were collected on 1
in the temperature range of 300-1.8 K in an applied field of 5
kG (Figure 2). The room temperature χMT value of approximately 18 cm3 K mol-1 remains constant until around 150
K, where it begins to increase to a maximum value of
approximately 36 cm3 K mol-1 at 20 K before dropping sharply
below this temperature. The spin only (g ) 2) value for a
[MnIV3MnIII4MnII2] unit is approximately 29 cm3 K mol-1. The
maximum in χMT is indicative of a high-spin ground state, with
the low temperature decrease assigned to zero-field splitting,
Zeeman effects, and/or intermolecular antiferromagnetic interactions. To determine the spin ground state for complex 1
magnetization data were collected in the ranges of 10-70 kG
and 1.8-4.0 K, and these are plotted as reduced magnetization
(M/NµB) versus H/T in Figure 2. For a complex entirely
populating the ground state and experiencing no zero-field
splitting, the observed isofield lines should superimpose and
saturate at a value (M/NµB) equal to gS. For 1, this is clearly
not the case and is suggestive of appreciable ZFS. The data
were fitted by a matrix-diagonalization method to a model that
assumes only the ground state is populated, includes axial zerofield splitting (DŜz2), and carries out a full powder average.20
The corresponding Hamiltonian is given by
Figure 2. (a) Plot of χMT versus T for complex 1 measured in the
temperature range of 1.8-300 K in a field of 5.0 kG. (b) Plot of reduced
magnetization (M/N µB) versus H/T for complex 1. The solid lines are fits
of the data to an S ) 17/2 state with g ) 1.97(3) and D ) -0.29 cm-1.
Data were measured at 1-7 T and 1.8-4 K.
Table 1. Crystallographic Data for 1
formula
M, g mol-1
crystal system
space group
a, Å
b, Å
c, Å
R, deg
β, deg
γ, deg
V, Å3
T, K
Z
Fcalc [g cm-3]
crystal shape and color
crystal size [mm]
µ [mm-1]
unique data
unique data, (I > 2σ(F))
R1, wR2
C52H80Mn9N6O35
1843.68
monoclinic
P21/n
13.745(3)
23.688(5)
23.259(5)
90
96.49(3)
90
7524(3)
150(2)
4
1.628
brown blocks
0.28 × 0.23 × 0.12
1.546
15403
13552
0.0376/0.0939
[MnIII4MnII2O6]4+ ring in the same manner. The remaining
coordination sites are occupied by two water molecules and three
pyridines. All of the Mn ions are in distorted octahedral
geometries, with their oxidation states assigned using charge
balance considerations, bond lengths, and bond valence sum
(BVS) calculations (Table 3). The Jahn-Teller elongations of
the Mn3+ ions all lie approximately perpendicular to the plane
of the [MnIII4MnII2O6]4+ ring.
Ĥ ) DŜz2 + gµBµ0ŜzHz
(1)
where D is the axial anisotropy, µB is the Bohr magneton, µ0 is
the vacuum permeability, Ŝz is the easy-axis spin operator, and
Hz is the applied field. The best fit gave S ) 17/2, g ) 1.97(3),
and D ) -0.29 cm-1. The ground state can be rationalized by
assuming an antiferromagnetic interaction between the ferromagnetically coupled Mn4+ ions in the [MnIV3O]10+ ring and
the Mn3+ and Mn2+ ions in the [MnIII4MnII2O6]4+ ring. This is
analogous to the situation in [Mn12O12(O2CR)16(H2O)x]n- where
the four ferromagnetically coupled Mn4+ ions couple antiferromagnetically with the surrounding eight Mn3+ ions resulting
in a net spin of S ) 10.1b
Ac Magnetic Susceptibility. Given the nature of S and the
sign and amplitude of D for complex 1, ac susceptibility studies
were undertaken to see if 1 acts as a single-molecule magnet.
When the rate at which the magnetization of a molecule relaxes
is close to that of an applied ac field, an out-of-phase ac
susceptibility signal (χM//) will be observed, in conjunction with
a concomitant decrease in the in-phase susceptibility (χM/T). At
low temperature where thermal energy is not sufficient to
overcome the barrier for relaxation, the magnetization of the
molecule cannot keep in-phase with the oscillating field. At the
temperature of the χM// versus T maximum, the relaxation rate
(1/τ) equals the ac angular frequency (ω ) 2πυ), and thus the
measurement of peak maxima at several different frequencies
provides 1/τ versus T data (since ω ) 1/τ) and the barrier to
relaxation can then be estimated from the Arrhenius equation.
Ac susceptibility studies were carried out in the 1.8-10.0 K
(20) MAGNET, Davidson, E. R.; Indiana University.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005 5575
Piligkos et al.
ARTICLES
Table 2. Selected Bond Lengths (Å) and Angles (deg) for 1
Mn(1)-O(2)
Mn(1)-O(3)
Mn(1)-O(1)
Mn(1)-O(3T)
Mn(1)-O(11)
Mn(1)-O(1T)
Mn(2)-O(5)
Mn(2)-O(4)
Mn(2)-O(1)
Mn(2)-O(2T)
Mn(2)-O(12)
Mn(2)-O(1T)
Mn(3)-O(6)
Mn(3)-O(7)
Mn(3)-O(1)
Mn(3)-O(13)
Mn(3)-O(2T)
Mn(3)-O(3T)
1.8139(18)
1.8595(18)
1.9046(17)
1.9197(18)
1.9309(18)
1.9470(18)
1.8146(18)
1.8700(18)
1.9014(17)
1.9278(18)
1.9300(18)
1.9426(17)
1.8024(18)
1.8069(18)
1.9255(18)
1.9343(19)
1.9883(18)
1.9916(17)
O(2)-Mn(1)-O(3)
O(2)-Mn(1)-O(1)
O(3)-Mn(1)-O(1)
O(2)-Mn(1)-O(3T)
O(3)-Mn(1)-O(3T)
O(1)-Mn(1)-O(3T)
O(2)-Mn(1)-O(11)
O(3)-Mn(1)-O(11)
O(1)-Mn(1)-O(11)
O(3T)-Mn(1)-O(11)
O(2)-Mn(1)-O(1T)
O(3)-Mn(1)-O(1T)
O(1)-Mn(1)-O(1T)
O(3T)-Mn(1)-O(1T)
O(11)-Mn(1)-O(1T)
O(7)-Mn(4)-O(2)
O(7)-Mn(4)-O(29)
O(2)-Mn(4)-O(29)
O(7)-Mn(4)-O(14)
O(2)-Mn(4)-O(14)
O(29)-Mn(4)-O(14)
O(7)-Mn(4)-O(1W)
O(2)-Mn(4)-O(1W)
O(29)-Mn(4)-O(1W)
O(14)-Mn(4)-O(1W)
O(7)-Mn(4)-N(1A)
O(2)-Mn(4)-N(1A)
O(29)-Mn(4)-N(1A)
O(14)-Mn(4)-N(1A)
O(1W)-Mn(4)-N(1A)
O(5)-Mn(7)-O(4)
O(5)-Mn(7)-O(17)
O(4)-Mn(7)-O(17)
O(5)-Mn(7)-O(26)
O(4)-Mn(7)-O(26)
O(17)-Mn(7)-O(26)
O(5)-Mn(7)-O(2W)
O(4)-Mn(7)-O(2W)
O(17)-Mn(7)-O(2W)
O(26)-Mn(7)-O(2W)
O(5)-Mn(7)-O(22)
O(4)-Mn(7)-O(22)
O(17)-Mn(7)-O(22)
O(26)-Mn(7)-O(22)
O(2W)-Mn(7)-O(22)
84.48(8)
94.97(8)
92.22(8)
93.85(8)
174.54(8)
82.74(8)
94.25(8)
94.47(8)
169.09(8)
90.84(8)
174.77(8)
90.76(8)
83.01(7)
90.69(7)
88.29(8)
93.84(7)
85.72(8)
179.43(8)
176.70(8)
83.23(8)
97.22(8)
94.35(8)
94.23(8)
85.45(10)
87.41(10)
88.96(8)
91.60(8)
88.76(10)
89.59(10)
173.09(10)
81.42(8)
96.38(8)
177.55(8)
171.33(8)
91.87(8)
90.21(8)
97.34(8)
95.53(8)
85.82(8)
88.73(9)
89.17(8)
89.99(8)
88.88(8)
85.32(9)
172.01(8)
Mn(4)-O(7)
Mn(4)-O(2)
Mn(4)-O(29)
Mn(4)-O(14)
Mn(4)-O(1W)
Mn(4)-N(1A)
Mn(5)-O(2)
Mn(5)-O(3)
Mn(5)-O(24)
Mn(5)-O(15)
Mn(5)-O(111)
Mn(5)-O(21)
Mn(6)-O(3)
Mn(6)-O(4)
Mn(6)-O(16)
Mn(6)-O(25)
Mn(6)-O(211)
Mn(6)-N(1B)
O(5)-Mn(2)-O(4)
O(5)-Mn(2)-O(1)
O(4)-Mn(2)-O(1)
O(5)-Mn(2)-O(2T)
O(4)-Mn(2)-O(2T)
O(1)-Mn(2)-O(2T)
O(5)-Mn(2)-O(12)
O(4)-Mn(2)-O(12)
O(1)-Mn(2)-O(12)
O(2T)-Mn(2)-O(12)
O(5)-Mn(2)-O(1T)
O(4)-Mn(2)-O(1T)
O(1)-Mn(2)-O(1T)
O(2T)-Mn(2)-O(1T)
O(12)-Mn(2)-O(1T)
O(2)-Mn(5)-O(3)
O(2)-Mn(5)-O(24)
O(3)-Mn(5)-O(24)
O(2)-Mn(5)-O(15)
O(3)-Mn(5)-O(15)
O(24)-Mn(5)-O(15)
O(2)-Mn(5)-O(111)
O(3)-Mn(5)-O(111)
O(24)-Mn(5)-O(111)
O(15)-Mn(5)-O(111)
O(2)-Mn(5)-O(21)
O(3)-Mn(5)-O(21)
O(24)-Mn(5)-O(21)
O(15)-Mn(5)-O(21)
O(111)-Mn(5)-O(21)
O(5)-Mn(8)-O(6)
O(5)-Mn(8)-O(18)
O(6)-Mn(8)-O(18)
O(5)-Mn(8)-O(27)
O(6)-Mn(8)-O(27)
O(18)-Mn(8)-O(27)
O(5)-Mn(8)-O(110)
O(6)-Mn(8)-O(110)
O(18)-Mn(8)-O(110)
O(27)-Mn(8)-O(110)
O(5)-Mn(8)-N(1C)
O(6)-Mn(8)-N(1C)
O(18)-Mn(8)-N(1C)
O(27)-Mn(8)-N(1C)
O(110)-Mn(8)-N(1C)
range in a 3.5 G field oscillating at frequencies up to 1000 Hz.
The in-phase (χM/T) and out-of-phase (χM//) signals for 1 are
shown in Figure 3. There is a frequency-dependent decrease in
the χM/T signal and a frequency-dependent out-of-phase signal
at T e 4 K. Both are strong indicators (but not definitive
evidence) of SMM behavior. For complex 1, the peak at 1000
Hz occurs at approximately 3.5 K. The data obtained were fit
to the Arrhenius equation to give an effective energy barrier
for the reorientation of the magnetization (Ueff) of 27 K, with
τ0 ) 5.0 × 10-8 s.
Single-Crystal Measurements. Hysteresis Studies. If complex 1 is an SMM, it should display hysteresis loops in
5576 J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005
2.0992(18)
2.1193(18)
2.150(2)
2.158(2)
2.217(3)
2.282(3)
1.8565(18)
1.9077(18)
1.9489(19)
1.978(2)
2.183(2)
2.224(2)
1.8822(19)
1.8840(18)
1.966(2)
1.9728(19)
2.233(2)
2.274(2)
83.77(8)
95.52(8)
93.35(8)
94.88(8)
175.11(8)
82.08(7)
94.06(8)
94.94(8)
167.94(8)
89.85(8)
174.49(8)
90.94(8)
83.21(7)
90.27(8)
87.91(7)
81.99(8)
95.17(8)
173.96(9)
173.90(8)
93.32(8)
89.12(8)
98.94(8)
96.54(8)
89.15(9)
85.43(9)
90.13(8)
86.71(8)
87.97(9)
85.68(9)
170.69(8)
95.84(7)
179.07(8)
84.46(8)
84.44(8)
178.40(10)
95.28(8)
93.51(8)
92.99(9)
87.34(9)
85.41(10)
87.34(8)
87.75(8)
91.80(10)
93.84(10)
178.80(9)
Mn(7)-O(5)
Mn(7)-O(4)
Mn(7)-O(17)
Mn(7)-O(26)
Mn(7)-O(2W)
Mn(7)-O(22)
Mn(8)-O(5)
Mn(8)-O(6)
Mn(8)-O(18)
Mn(8)-O(27)
Mn(8)-O(110)
Mn(8)-N(1C)
Mn(9)-O(6)
Mn(9)-O(7)
Mn(9)-O(28)
Mn(9)-O(19)
Mn(9)-O(210)
Mn(9)-O(23)
O(6)-Mn(3)-O(7)
O(6)-Mn(3)-O(1)
O(7)-Mn(3)-O(1)
O(6)-Mn(3)-O(13)
O(7)-Mn(3)-O(13)
O(1)-Mn(3)-O(13)
O(6)-Mn(3)-O(2T)
O(7)-Mn(3)-O(2T)
O(1)-Mn(3)-O(2T)
O(13)-Mn(3)-O(2T)
O(6)-Mn(3)-O(3T)
O(7)-Mn(3)-O(3T)
O(1)-Mn(3)-O(3T)
O(13)-Mn(3)-O(3T)
O(2T)-Mn(3)-O(3T)
O(3)-Mn(6)-O(4)
O(3)-Mn(6)-O(16)
O(4)-Mn(6)-O(16)
O(3)-Mn(6)-O(25)
O(4)-Mn(6)-O(25)
O(16)-Mn(6)-O(25)
O(3)-Mn(6)-O(211)
O(4)-Mn(6)-O(211)
O(16)-Mn(6)-O(211)
O(25)-Mn(6)-O(211)
O(3)-Mn(6)-N(1B)
O(4)-Mn(6)-N(1B)
O(16)-Mn(6)-N(1B)
O(25)-Mn(6)-N(1B)
O(211)-Mn(6)-N(1B)
O(6)-Mn(9)-O(7)
O(6)-Mn(9)-O(28)
O(7)-Mn(9)-O(28)
O(6)-Mn(9)-O(19)
O(7)-Mn(9)-O(19)
O(28)-Mn(9)-O(19)
O(6)-Mn(9)-O(210)
O(7)-Mn(9)-O(210)
O(28)-Mn(9)-O(210)
O(19)-Mn(9)-O(210)
O(6)-Mn(9)-O(23)
O(7)-Mn(9)-O(23)
O(28)-Mn(9)-O(23)
O(19)-Mn(9)-O(23)
O(210)-Mn(9)-O(23)
1.8605(18)
1.9111(18)
1.9623(19)
1.983(2)
2.197(2)
2.219(2)
2.1142(18)
2.1159(19)
2.132(2)
2.169(2)
2.221(2)
2.281(3)
1.8755(18)
1.8779(18)
1.9778(19)
1.9837(19)
2.197(2)
2.237(2)
84.89(8)
96.03(8)
94.93(8)
95.00(8)
94.18(8)
166.26(8)
92.58(8)
174.02(8)
79.93(7)
91.43(8)
175.94(8)
93.55(8)
80.35(7)
88.85(8)
88.61(7)
96.47(8)
174.07(8)
88.89(8)
89.48(8)
173.57(8)
85.06(8)
93.93(9)
98.14(8)
87.80(9)
83.83(8)
90.31(8)
91.54(8)
86.99(9)
85.98(8)
168.92(8)
80.93(7)
94.54(8)
172.27(8)
174.65(8)
94.62(8)
89.55(8)
97.68(8)
97.46(8)
89.32(9)
85.79(8)
89.36(8)
86.76(8)
86.93(9)
87.42(8)
172.26(8)
magnetization versus field studies. Figure 4 shows just such a
plot for 1, at temperatures below 1.5 K and at a sweep rate of
0.001 T s-1 and at 1.05 K at sweep rates of 0.001-0.017 T
s-1, with the field applied in the direction of the easy-axis of
half the molecules. For the other half, the field is approximately
transverse. The hysteresis loops are sweep rate and temperaturedependent and are not smooth, showing steps at regular intervals
of field indicative of quantum tunneling (QTM). The field
separation between the zero-field resonance and the next for 1
is 0.27 T, giving D/g ) 0.129 cm-1. If we assume a value of g
) 2, then the observed D ) -0.258 cm-1 is in good agreement
with that obtained from the bulk dc measurements.
Enneanuclear Manganese Single-Molecule Magnet
ARTICLES
Table 3. Bond Valence Sum (BVS) Calculations for the Metal
Ions in 1
Mn1
Mn2
Mn3
Mn4
Mn5
Mn6
Mn7
Mn8
Mn9
MnII
MnIII
MnIV
assignment
4.109
4.117
4.034
1.965
3.140
3.039
3.090
1.958
3.082
4.188
4.196
4.111
2.003
3.200
3.097
3.149
1.995
3.141
4.542
4.550
4.458
2.172
3.471
3.359
3.415
2.164
3.407
MnIV
MnIV
MnIV
MnII
MnIII
MnIII
MnIII
MnII
MnIII
rate of 8 × 106 s. In this temperature regime, the relaxation of
the magnetization occurs only via quantum tunneling as the
thermally activated process becomes insignificant.
Inelastic Neutron Scattering (INS) Study of Undeuterated
1. Inelastic neutron scattering (INS) is a powerful tool to
determine exchange couplings and single-ion anisotropies in
molecular magnets.21 Due to the spin of the neutron, there is a
strong interaction between the magnetic moment of the neutron
and the magnetic field created by the unpaired electrons in the
sample. Measuring the INS intensity as a function of momentum
and energy transfer provides useful information on the magnetic
scattering function, S(Q,ω), which contains the details of the
magnetic exchange interactions and anisotropy splittings.22 Thus
far, none of the above experimental studies has provided direct
information on the nature and energies of energy splitting in
the ground state of 1. As a first approximation, one can use a
giant spin model to describe the ground state, assuming that it
is well isolated from the excited spin states. An effective
Hamiltonian then describes the zero-field splitting of the giant
spin ground state. The low symmetry of complex 1 allows for
many higher order crystal field terms in the effective spin
Hamiltonian. The fact that the Jahn-Teller axes of the four
MnIII ions all lie roughly perpendicular to the plane of the
[MnIII4MnII2O6]4+ ring and the presence of a pseudo 3-fold axis
perpendicular to this ring justifies the use of an axial Hamiltonian in the first instance:
Haxial ) D[Ŝ/z2-S(S + 1)/3] + B40Ô40
(2)
where
Ô40 ) 35Ŝz4 - [30S(S + 1) - 25]Ŝz2 - 6S(S + 1) +
3S2(S + 1)2
The first term in eq 2 splits the ground state into S + 1/2
Kramer’s doublets, and for a negative D value, the MS ) (S
doublet lies lowest in energy. Equation 2, therefore, splits this
ground state into nine (MS Kramer’s doublets. The INS
selection rule, ∆MS ) (1, predicts eight transitions for an S )
17/ ground state where the intensity for an |M | f |M + 1|
2
S
S
INS transition can be calculated as
Figure 3. Plots of the in-phase (χM/) signal as χM/T and out-of-phase (χM//)
signal in ac susceptibility studies versus temperature in a 3.5 G field
oscillating at the indicated frequencies.
Relaxation Studies using Ac and Dc Data. Relaxation data
were also determined for 1 from dc relaxation decay measurements performed on single crystals using an array of microSQUIDS. First, a large dc field of 1.4 T was applied to each
sample at 5 K to saturate the magnetization in one direction,
and the temperature decreased to a specific value between 1.8
and 0.04 K. When the temperature was stable, the field was
swept from 1.4 to 0 T at a rate of 0.14 T s-1 and the
magnetization in zero-field measured as a function of time. A
plot of the relaxation, plotted as a fraction of maximum value
MS, for complex 1 is shown in Figure 4. Combining this
relaxation data with the ac data allows the construction of an
Arrhenius plot of τ versus 1/T. Above approximately 0.4 K the
relaxation rate is temperature-dependent with the fits yielding
Ueff ) 26.6 K and τ0 ) 6.5 × 10-8 s. Below 0.4 K, however,
the relaxation rate is temperature-independent with a relaxation
I(|MS| f |MS + 1|) ∝ P((MS) ×
S
1 S
S
1 S
MS
-MS 1
MS + 1 × -MS - 1 1
(
) (
)
(3)
where P((MS) is the Boltzmann population of the initial state
(MS.
Figure 5 shows the INS spectra of 1 measured at 1.6 (10 h),
12 (31 h), and 20 K (25 h), summed over all scattering angles.
Positive and negative energy transfer corresponds to neutron
energy loss and gain, respectively. The 1.6 K spectrum exhibits
one strong band at 3.72 cm-1 on the loss side with a second
one coming up as a shoulder at lower energy. A series of hot
bands is observed at 12 and 20 K. At this latter temperature,
we measured the energy gain and the loss side. The peak
energies are the same on both sides, and we can identify six
(21) Basler, R.; Boskovic, C.; Chaboussant, G.; Güdel, H.-U.; Murrie, M.;
Ochsenbein, S. T.; Sieber, A. ChemPhysChem 2003, 4, 910-926.
(22) Böni, P.; Furrer, A.; Schefer, J. Principles of Neutron Scattering, Proceedings of the Eigth Summer School on Neutron Scattering, World Scientific
Co. Pte. Ltd., 2000.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005 5577
Piligkos et al.
ARTICLES
Figure 4. Magnetization (M) of 1 (plotted as fraction of maximum value MS) versus applied magnetic field (µ0H). The resulting hysteresis loops are shown
at different temperatures (a) and different field sweep rates (b). Single-crystal relaxation data for 1, plotted as fraction of maximum value MS versus time
(c). Arrhenius plot for 1 using ac data (χM//) and dc decay data on a single crystal. The solid line is the fit of the data in the thermally activated region, and
the dashed line is the fit of the temperature-independent data.
Figure 5. (a) INS spectra measured on IRIS with an analyzing energy of 1.84 meV at 1.6 (10 h), 12 (31 h) and 20 K (25 h), summed over all scattering
angles. (b) INS spectrum of 1 measured at 20 K corrected for background (top) and calculated spectrum at 20 K using eqs 2 and 3 with the parameters D
) -0.249 cm-1 and B40 ) 7 × 10-6 cm-1 (bottom). (c) Calculated anisotropy splitting of the S ) 17/2 ground state according to the Hamiltonian of eq 2
with parameter values of D ) -0.249 cm-1 and B40 ) 7 × 10-6 cm-1. The arrows correspond to the allowed ∆MS ) (1 INS transitions.
distinct maxima and a shoulder on the elastic line on either side
of the spectrum. The peaks exhibit an almost regular picket fence
pattern with decreasing peak separations toward higher energy
5578 J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005
transfer. On the basis of their temperature and Q-dependence,
it is possible to assign the 14 observed bands to ∆MS ) (1
transitions within the ground state multiplet. The progression
Enneanuclear Manganese Single-Molecule Magnet
of the peak positions is limiting the possible number of
transitions in the ground state multiplet to exactly eight on either
side.
Figure 5b shows the spectrum measured at 20 K after
subtracting the nonmagnetic background. The peaks labeled
I-VII correspond to the transitions shown in Figure 5c, which
shows the energy splitting pattern of the S ) 17/2 ground state.
Least-squares calculations using eq 2 and the peak positions
determined by INS yield the following best parameter set (D
) -0.249(5) cm-1, and B40 ) 7(4) × 10-6 cm-1. The value of
the axial anisotropy parameter, D, is in good agreement with
the D values of -0.29 and -0.26 cm-1 derived from powder
and single-crystal magnetic measurements, respectively. The use
of a Hamiltonian without the fourth-order term, B40Ô40, could
not account for the observed decrease of the energy intervals
with increasing energy transfer. The B40 parameter must be
positive, which is not without precedent in the spin cluster
literature (positive B40 values have been reported for the singlemolecule magnets [Fe8] and [Mn4]).23,24
Figure 5 shows a calculated spectrum at 20 K using eqs 2
and 3 with the best fit parameters. The peak positions and
relative intensities are in excellent agreement with the experimental data. For energy transfers below (1 cm-1, the experimental data are dominated by the elastic line, and the eighth
peak predicted by the calculations is not observable. The
nonmagnetic scattering is orders of magnitude more intense than
the magnetic scattering in the region of the elastic line, thus
swamping the magnetic part. Physically, the strong central line
with the pronounced tails in Figure 5a originates from elastic
and quasi-elastic nonmagnetic scattering processes. Its intensity
is enhanced by the presence of a large number of hydrogen
atoms in the measured sample. At the low energy transfer of
our experiment, it leads to an enhanced elastic line, including
tails as well as a higher background. For a more accurate
determination of peak VII in Figure 5b, it would be necessary
to measure a deuterated sample. The incoherent nuclear scattering cross-section of deuterium is 40 times smaller than that
of hydrogen; therefore, the effects described above are reduced
drastically. Deuteration is often considered a prerequisite for a
successful magnetic INS experiment, but the present study
shows that data of remarkable quality can be obtained with an
undeuterated sample in the energy transfer range between 1 and
4 cm-1. The energy transfer region, in which magnetic cluster
excitations can be observed despite the presence of hydrogen
atoms, covers the range from about 1 to 30 cm-1. We ascribe
this to a low density of vibrational states with significant
hydrogen contribution in this “lucky window”.
Frequency Domain Magnetic Resonance Spectroscopy
(FDMRS) of 1. FDMRS spectra were recorded on a polycrystalline powder pellet of complex 1 using linearly polarized
radiation in the frequency range of 2.8-5.0 cm-1 and at
temperatures between 1.8 and 30 K, in the absence of an external
magnetic field. The spectra were normalized, that is, divided
by the 30 K spectrum to remove the oscillation of the baseline
caused by Fabry-Perot-like interference within the plane
parallel sample as well as artifacts caused by standing waves.
(23) Caciuffo, R.; Amoretti, G.; Murani, A.; Sessoli, R.; Caneschi, A.; Gatteschi,
D. Phys. ReV. Lett. 1998, 81, 4744-4747.
(24) Yoo, J.; Yamaguchi, A.; Nakano, M.; Krzystek, J.; Streib, W. E.; Brunel,
L.-C.; Ishimoto, H.; Christou, G.; Hendrickson, D. N. Inorg. Chem. 2001,
40, 4604-4616.
ARTICLES
Figure 6. Normalized FDMRS spectra recorded on a pressed powder pellet
of 1 at various temperatures, as indicated in the figure. The inset shows the
1.8 K non-normalized spectrum together with fits assuming Gaussian (solid
line) and Lorentzian (dotted line) line shape functions.
In the normalized spectra (Figure 6), three resonance lines can
be observed at ν ) 3.79, 3.41, and 2.95 cm-1. The exact
positions of the resonance lines were determined by fitting them
with high degree polynomials. The obtained resonance frequencies were then used in the fitting procedure. From the fact that
the highest frequency resonance line has highest intensity at
the lowest temperatures, it can immediately be concluded that
the D term in the zero-field splitting Hamiltonian must be
negative. A first value of this parameter was determined by
searching the minimum discrepancy between calculated and
observed resonance frequencies. Introducing an E term and
minimizing the discrepancy factor in two dimensions leads to
a value that is practically zero and did not significantly decrease
the discrepancy factor. A significantly better fit is obtained when
a B40 term is included. Adding transverse anisotropy in fourth
order (a B43 term in the case of trigonal symmetry) also did not
improve the fit. Finally, the best fit parameters were determined
to be D ) -0.247 ( 0.005 cm-1 and B40 ) (+4.6 ( 0.1) ×
10-6 cm-1. This means that the system is completely axial and
that the MS states are the eigenstates of the system. The observed
resonance lines correspond, therefore, to the transitions |MS )
(17/2> f |MS ) (15/2> (3.79 cm-1), |MS ) (15/2> f
|MS ) (13/2> (3.41 cm-1), and |MS ) (13/2> f |MS )
(11/2> (2.95 cm-1).
From the obtained spin Hamiltonian parameters, the spectra
can be simulated. To this end, the magnetic permeability of the
sample calculated from the sum of the contributions due to the
various resonance modes and the dielectric permeability obtained
from the oscillations of the baseline and the thickness of the
sample were used. The transmission spectrum is calculated from
them using Fresnel’s formulas.25 The inset in Figure 6 shows
that the spectrum can be very well simulated. The simulation
assuming a Gaussian line shape is slightly better than that
assuming a Lorentzian one, but definitive conclusions are
difficult to draw. Indeed, if line broadening contributions due
to both homogeneous and inhomogeneous processes are of the
same order, the line shape might be intermediate between these
two extremes.
Theoretical Studies: DFT Calculations. Using the structural
model described in the Experimental Section, we carried out
(25) Dressel, M.; Grüner, G. Electrodynamics of Solids; Cambridge University
Press: Cambridge, U.K., 2002.
J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005 5579
Piligkos et al.
ARTICLES
Table 4. Comparison of zfs Parameters Obtained by FDMRS,
INS, Reduced Magnetization versus H/T Measurements, and
Single-Crystal Measurements on 1
zfs
(cm-1)
D
B40
Figure 7. Plot of the spin density obtained from first-principles DFT
calculations. Red corresponds to “spin up” and blue to “spin down”.
DFT calculations at the all-electron level on a single-molecular
entity. During the self-consistent cycle, we did not apply any
constraint on the total moment of the cluster, allowing a search
of the full space of magnetic moments. The calculation
converged to a ground state with 17 unpaired electrons, in
agreement with the experimental findings. The magnetic ordering is shown in Figure 7 by a plot of the spin density.
To identify the charge state of the Mn ions, we calculated
the magnetic moment within a sphere of radius 1.2 Å centered
at the Mn site. For the three inner Mn(IV) ions with “spin down”
electrons, we obtained a magnetic moment of -2.4 µB. The
outer ring of Mn ions all have “spin up” electrons with two
Mn ions displaying 4.3 µB and the remaining Mn ions having
values of 3.3, 3.5 (×2), and 3.6 µB. Considering that this method
of calculating the magnetic moment will not include all
contributions to the spin density of a given atom, one can
conclude that the closest ionic picture is fully consistent with
the earlier experimental findings of two Mn(II) and four Mn(III) ions being spin up and three Mn(IV) ions being spin down,
resulting in the S ) 17/2 spin ground state of the molecule.
Using self-consistent wave functions18,19 allows us to take
into account all second-order contributions to the magnetic
anisotropy due to spin-orbit interactions. These contributions
can be expressed in the form of a simple spin Hamiltonian with
only quadratic terms in the spin operators:
H ) DSz2 + E (Sx2 - Sy2)
(6)
The rhombohedral anisotropy parameter, E, vanishes for uniaxial
systems due to symmetry, but for lower symmetry systems, such
as the one considered here, it may have a nonzero value. The
final results obtained from the NRLMOL first-principles DFT
calculation are D ) -0.23 cm-1 and E ) 0.035 cm-1.
Therefore, the system is an easy-axis system with a magnetic
anisotropy barrier of 26.2 K. The value of the axial D is in
very good agreement with the experimental observation and of
5580 J. AM. CHEM. SOC.
9
VOL. 127, NO. 15, 2005
FDMRS
INS
-0.247 ( 0.005
-0.249(5)
(+4.6 ( 0.1) × 10-6 +7(4) × 10-6
magnetization µ-SQUID
-0.29(3)
DFT
-0.258 -0.235
similarly good quantitative agreement with values found for
other single-molecule magnets.19
We would like to stress here that there is no experimental
input besides the molecular structure and no free adjustable
parameter. The numerical accuracy of the calculated value of
D with respect to the basis set and other purely numerical
parameters is estimated to be of the order of 0.01 cm-1, and
similar changes are expected for small changes in the geometry
of the cluster. However, this error estimate does not include
systematic deviations due to the approximate nature of DFT,
which may result in a D value smaller by a factor of 2 compared
with that of experiment.26 The coupling of the vibrational
degrees of freedom to the spin may give rise to fourth-order
contributions to the magnetic anisotropy energy. While in
principle it is possible to calculate these contributions,27 we have
not attempted this because it was computationally too expensive
with our currently available resources. However, considering
the small value of the fourth-order contribution indicated from
experiment, our numerical accuracy may not be sufficient for
such an estimate.
Conclusions
Reaction of [Mn3O(O2CMe)6(py)3] with the tripodal ligand
H3thme (1,1,1-tris(hydroxymethyl)ethane) produces the complex
[Mn9O7(O2CCH3)11(thme)(py)3(H2O)2] whose mixed-valent metallic skeleton describes part of an idealized icosahedron.
Magnetic studies using a variety of techniques, including bulk
and single-crystal magnetometry, inelastic neutron scattering,
frequency domain magnetic resonance spectroscopy, and theoretical density functional calculations, show the complex to
possess a spin ground state of S ) 17/2. The ground state arises
from the antiferromagnetic coupling between the central Mn4+
ions and the peripheral Mn3+ and Mn2+ ions. Hysteresis loops
at ultralow temperatures show the complex to be a singlemolecule magnet with an energy barrier of approximately 27
K.
Acknowledgment. We wish to thank Grégory Chaboussant,
Felix Fernandez-Alonso, and Stefan T. Ochsenbein for assistance
in the collection of INS data. We also thank Lloyd’s of London
Tercentenary Foundation and the EPSRC (UK) for funding.
Supporting Information Available: Cartesian coordinates
used for the DFT calculation, and complete refs 5, 8d, and 13.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA042302X
(26) Cirera, J.; Kortus, J.; Ruiz, E. To be published.
(27) Pederson, M. R.; Bernstein, N.; Kortus, J. Phys. ReV. Lett. 2002, 89,
097202-1/4.